I have been fascinated by microexons for a long time. Many exons are relatively small (less than 100 bp.), but still large enough for recognition of the two splice sites by the splicing machinery simultaneously. In fact, the boundaries of such exons are often recognized coordinately in a process known as exon definition. Introns that are too small for that, so that the two splice sites cannot be recognized simultaneously, are termed microexons. Many are less than 10 nucleotides (see Volfovsky et al. 2003). Sometimes the downstream intron must be removed first (e.g. potato invertase). The inability of the splicing machinery to recognize both splice sites simultaneously due to physical occlusion probably comes into play with exons less than about 30 nucleotides. Although the exact length at which this occurs is difficult to know for sure (especially in a species like S. mansoni, for which we have little experimental data, it is nevertheless what I consider to be the defining characteristic of a true microexon. Microexons are characterized by alternative splicing and annotation errors.
Now, the genome of the blood fluke Schistosoma mansoni reveals "at least 45 genes with an unusual microexon structure," such that microexons make up the majority of the coding sequence in those genes. As is often true with microexons, these genes are alternatively spliced, suggesting that a "'pick and mix' strategy is used to create protein variation." These MEGs, or microexon genes, have the hallmarks of secreted proteins and are expressed in the intramammalian stages of the life cycle. It will be interesting to see what role microexon splicing or these genes turns out to play.